Topic: CoinTerra announces its first ASIC - Hash-Rate greater than 500 GH/s - page 87. (Read 231002 times)

I'm quite sure the fab works every day and doesn't take off weekends.

Also when you talk of "days" in this industry are they working days or full week days including weekends and holidays?

That is a good question.

BTW..


One has to wonder how they ever expected to ship miners a few weeks before receiving wafers. Didnt they always claim the silion production was on schedule?

keep in mind a bitcoin asic almost certainly requires less metal layers than your typical soc/cpu/gpu, which will decrease fabrication time proportionally. 24 days sounds very fast to me too, but I doubt that number came out of nowhere. I actually suspect its the fabrication time, not including the tape out process and mask generation, but FWIW.

Bullet run    Rocket run   Hot Lot.   45 days at best.   it does not make each layer go faster, it reduces the time you are waiting between each layer.

According to BFL, a "bullet run" at globalfoundries takes 24 days. To quote Josh "The initial process is what is called a bullet run, it will allow us to get a finished set of wafers in approximately 24 days from the date we tape out."  At TSMC afaik hot lots take longer than that. Does anyone know where CT is fabbing?

As for what BFL actually achieved, if you are referring to their 65nm chip, Im quite sure they failed their first tape out attempt (the "refraction issue"), and had to redo their physical design flow. I dont think they told us when they successfully taped out.

there's 22 days of nov and 31 days of dec left this year, 53 days total.  its certainly possible (but aggressive) to be done in time.  depends on the tape-out date.   But i don't see January as a risk.. that adds another 31 days on top of the 53...  if they're going to tape-out imminently, the dec dates are aggressive but not impossible, and the jan deliveries aren't aggressive at all.

now that hashfast has announced deliveries starting dec 15th onwards... it will be interesting to see when cointerra ships, and whether they will manage to eek back any of the time difference between their schedules.  certainly in tape-out, hashfast is more than 2 months ahead... but if cointerra can execute in anywhere near the speed that knc did, that delta between the two won't be as pronounced.  only time will tell.  meanwhile, we await cointerra's tape-out announcement with baited breath...

And anyway - this isn't a race between hf and cointerra, as they are different products at different price points... and cointerra's not announced any delay (...yet!).  they have schedules for dec and jan deliveries, and so far at least, they haven't announced any delays...

I can't seem to find KNC tapeout announcement.  I believe they announced after the fact but I can't seem to find even the belated announcement anymore.   Still say KNC was 45 days.

That means with three vendors:
KNC = 45 days
Hashfast = 70 days.
BFL >70 days (still not provided any detail on exactly when taped out and hasn't yet indicated they received chips unless they were lying about Aug tapeout it is probably >70 days).

Throw another two weeks for post fabrication work and a week to hedge for any issue and you are looking at 60-90 days (possibly more for BFL) from tapeout to shipping depending on how perfect and how fast the fabrication is.  

Given that I can't really see how Cointerra is going to ship in Dec (31 DEC is 54 days away) and even Jan (31 JAN is 85 days away) becomes increasingly hard to believe if they haven't taped yet.  Maybe Cointerra would like to show us how they are going to meet their timeline or provide an update pushing shipping schedule back.  At this point they have to know if Dec/Jan is realistic or impossible waiting until Dec is just bad faith.   It isn't like they don't know.  They know if the tapeout is done, they know their delivery window from the fab. 

neither bfl nor hashfast broke any records in bitcoin mining asics from tape-out to production... i suspect that honour goes to kncminer.

both bfl and hf said they taped out in august.  hf was near the end of august... thus bfl could in theory have been earlier (but not much later).  thus had bfl paid for expedite fees, they would've had silicon by now...

its my understanding that kncminer, from tapeout to shipping boxes to customers was 60 days !  and if you work backwards from the 60 days... they had time to get their 4 dies mounted on the substrates, manufacture the pcbs, assemble the boxes (bring up the boards, get the firmware working) etc...  (knc fabbed their chips at tsmc as well so the timings could've been similar).  clearly, knc mustve paid extra for an expedited fab and also had all their ducks in a row for everything else they needed that wasn't just the silicon.. and that, it seems, is why knc was able to ship so quickly and has caused issues for hf.   More likely knc's was 40-45 ish days from tape-out to silicon, leaving them 15-20 days for bumping, dies onto substrate, boards, and the boxes...  

Actually.. being specific... 28nm asics have about 50 mask layers.. and takes around a day to make each mask layer.  plus some extra time at the end for bumping etc.    you pay extra to get each layer done a little faster..  you pay a lot extra for 0.8 or 0.7 Days Per ML.  You pay a HUUUGE amount extra for 0.6 or maybe even faster DPML.  in theory the fastest you could get 50 layers made is probably about 0.5days/ML (i.e.: 25 days in fab plus a few days for bumping etc).  lets assume 30 days is the fastest it could be done, and 40 is more reasonable.

So for those looking for a second reference point.  Hashfast announced today they will be delayed at least three to four weeks from today (ouch) but kinda buried in that was the fact that they received the final silicon today.

Hashfast announces tapeout complete = 08/28/2013.
Hashfast announces they have final silicon = 11/07/2013.
Fabrication took 71 days from tapeout. 

That is inline with BFL 10 week time line.

No problem(as long as you keep up with the good posts) and yes same here. I would buy their product once they have shown they can deliver.

They announced the mock tape-out and their sales would benefit from the certainty provided by the real one.

We don't know for sure.  But presumption is negative, so we presume that have not until they tell us otherwise.

Sorry nothing personal I get a LOT of PMs a day and some days I don't get to them all then they get buried and I forget about them.

I would point out everything is based on probable timelines and the fact that I (just my opinion) find it highly unlikely that IF Cointerra already taped out weeks ago they would just remain silent.  Press releases are cheap and a confirmed tapeout would spur MORE sales.  I don't think Cointerra is a "scam", I don't think made up a website to run off with the money but I do believe (for whatever reason) they haven't completed the tapeout YET.  Most likely they will and most likely they will eventually ship but one can count on shipping being at least 60 (being optimistic again) days from tapeout.  You can't push the tapeout back 30, 40, 50, 60 days without shipping also being pushed back.  I got nothing against Cointerra, I may buy their product in the future (once they have shown they can deliver) and if the price is right.

I was ready for a purchase too, but thankfully you not replying to my PM about this made me reconsider

There is no one exact timeline but it is roughly accurate.  I really (based on past experiences) doubt BFL will be able to meet their own timeline but the timeline is realistic (maybe a little optimistic but not impossible) for a competent company.

BFL's schedule is more or less realistic (if executed by someone else ).
Wafer production can be accelerated if you pay for it. BFL themselves claim to do that for monarch, reducing the time to 25 days (according to them).
Cointerra is also rumoured to have paid for this for their first batch, no idea about HF, but its irrelevant since their wafers are already done.

2 weeks for packaging seems very long to me, unless they send the wafers by slow boat from Germany to Asia and get the packaged chips back to the US. The actual packaging process itself is almost instantaneous.

As for the FUD about their experience. I really have to laugh at that.
Here is my interpretation: "We fucked up the first tape out attempt because we didnt have a clue and we had never done full custom asic design. We learned a few new words in the process, like "refraction". Then we fucked up our packaging choice because we fucked up our power consumption estimates and we just ordered the cheapest possible packaging. Even when we selected the correct package type appropriate for the power consumption, we still managed to fuck up the underfill. And once we finally got the chip to work, we had all sorts of issues with the PCB, cooling, firmware. It took a year longer than expected, but now we are the experts because we made every mistake there is to make in this industry".

Makes me chuckle. As KnC and many others proved, if you know what you are doing, none of those are hurdles one expects to bump in to. Bumping in to all of them just proves BFL hired completely incompetent engineers or design house.


Thats always possible, but unlikely, and nothing suggests this is this case, and unless they meanwhile solved the substrate sourcing issue, they wouldnt even know yet.
FWIW, even BFL didnt need a respin to get their asic working.

This is the timetable according to BFL after a tape-out.


Is this accurate or are they just trying to scare away competition and customers that go with the competition?

How long did it take KnC & BitFury to ship product starting from tape-out to start the process over?

Is it also possible that HashFast's chips aren't working and they have to go back to the drawing board for another tape-out?

If they taped out a month ago that would seem probable.  However having not taped yet, Dec becomes highly unlikely (I would say impossible but like cracking a Bitcoin private key I guess it is possible) and even Jan is questionable.  

If we assume they pay for a high cost low volume "rocket run" you are still looking at 30-45 days from tapeout before wafers are fabricated.  That assumes no delays or complications at the foundry.  The foundry will pad their windows with another 2-3 weeks in the contract so they can juggle around orders to keep the fab at 100% capacity.  Remember that fab cost the company a couple billion and the value of the wafers it produces falls with Moore's law.  Downtime is not an option and if necessary they can (and routinely do) bump small orders to reoptimize their operation.

So say Cointerra taped out today, wafers might be done in early to late Dec.  Add in 7-10 days for wafer transport, cutting, packaging, and testing and you are looking at late Dec, early Jan before they have chips.  Remember a wafer isn't a chip so fabrication taking 45 days doesn't mean on the 45th day you have a chip ready to solder to a board.  You have a giant wafer which needs to be cut, tested, packaged, and shipped.  Even if they are amazing and have all their ducks in a row give it a week for all PCB assembly, production, testing, and shipping.  That puts shipping starting in early to mid Jan (and I am being optimistic).  At this point (with tapeout not done) there is no slack in the schedule.  It isn't like budgeting 8 weeks for chips but it doesn't matter if you are 5 weeks late starting because you can get chips in 3 weeks.  Any delay starts pushing you into Feb range.  Without tapeout done there is no way they can be further along in the schedule.  Packaging can't happen until you have wafers, wafers can't start until tapeout is done.

I almost bought in early Oct but the lack of a tapeout announcement concerned me (not bragging I bought from three companies and none have arrived yet).  Even with a tapeout on Oct 1st, a Dec delivery would be pushing the limit. Now consider that we are 38 days past that point and the tapeout isn't done.

I do believe its extremely unethical to change that shipping date to february without telling us why.  Is it the same batch?  Is it a different batch?  We may never know.. very unprofessional

I'm also worried about how expensive it would be to have them host our machines.  Does this mean the costs would be higher if I do it myself?     $400/month seems outrageous